autoevolution
 

Waymo Forced To Recall Its Robotaxis Because They Kept Hitting Telephone Poles

Waymo forced to recall its robotaxis because they kept hitting telephone poles 6 photos
Photo: 12news.com
Waymo forced to recall its robotaxis because they kept hitting telephone polesWaymo forced to recall its robotaxis because they kept hitting telephone polesWaymo forced to recall its robotaxis because they kept hitting telephone polesWaymo forced to recall its robotaxis because they kept hitting telephone polesWaymo forced to recall its robotaxis because they kept hitting telephone poles
Waymo was forced to issue its second recall ever after one of its robotaxis in Phoenix, Arizona, hit a telephone pole, and another car stuck behind it moments later. The problem showed that relying on high-definition maps for autonomous driving might not pan out. This gives Tesla's vision-only alternative an unexpected boost.
There are roughly two trends in autonomous driving, and one appears to be losing the game. Most companies involved in vehicle autonomy believe that adding more sensors of various types improves accuracy and safety. Robotaxi companies are the best example, with their cars crammed with massive sensor arrays comprising radars, lidars, cameras, and ultrasonic sensors. These companies rely on high-definition map data to offer precision positioning.

The other trend has been pioneered by Tesla and its vision-only systems, which only use cameras to navigate the world, just like humans do. Tesla argued that adding more sensors would only contribute more noise, making it more difficult for neural networks to calculate the vehicle's path and behavior. Tesla only uses cameras, similar to how humans rely on their eyes for the same purpose.

For years, the former crowd has mocked Tesla for its strategy, especially as the company's Full Self-Driving software didn't appear to be making progress. However, Tesla's approach might prove better in the end, as relying on precision maps could make vehicles vulnerable to unexpected changes in the environment. Updating the maps takes time, and frequent roadworks can make this futile.

This was made clear after one of Waymo's driverless cars struck a telephone pole at the end of May. The car had no passengers inside, but it was driving to a passenger pickup location. The funny thing is that the crash happened close to the customer waiting for the ride, who heard the loud bang. Jericka Mitchell was offered a refund and a new ride, but the second car also got stuck behind the wreck of the first one.

The alley where the accident happened is lined up with wooden telephone poles on both sides. They are not up on a curb but at the road level, which might have contributed to the crash. In a statement, Waymo explained that its self-driving software incorrectly assigned a low damage score to the telephone pole. Waymo issued a recall of 672 vehicles to update the software and mapping data to account for the hard road edge in that particular alleyway.

Waymo's problem was quickly picked up by Elon Musk, who explained that relying on high-definition maps makes these vehicles prone to mapping errors. As the world is constantly changing, HD maps cannot be updated quickly enough to account for these changes. Another issue is that HD mapping data is not available for every street, and scaling it worldwide is very difficult.

This is Waymo's only second recall in its history after the one in February when two of its autonomous vehicles crashed in quick succession because of a software error. The recall comes at a time of intense regulatory scrutiny of the autonomous driving industry. Waymo, in particular, is subject to an NHTSA investigation after over two dozen incidents involving its robotaxis. These included single-party crashes, traffic violations, and crashes with stationary objects like the one with the telephone pole.

If you liked the article, please follow us:  Google News icon Google News Youtube Instagram
About the author: Cristian Agatie
Cristian Agatie profile photo

After his childhood dream of becoming a "tractor operator" didn't pan out, Cristian turned to journalism, first in print and later moving to online media. His top interests are electric vehicles and new energy solutions.
Full profile

 

Would you like AUTOEVOLUTION to send you notifications?

You will only receive our top stories